Estimation of glomerular filtration rate (eGFR) is one of the current clinical methods for identifying risk for diabetic nephropathy in subjects with type 1 diabetes (T1D). Hyperglycemia is known to influence GFR in T1D and variability in blood glucose at the time of eGFR measurement could introduce bias in eGFR. We hypothesized that simultaneously measured blood glucose would influence eGFR in adults with T1D.
Longitudinal multivariable mixed-models were employed to investigate the relationships between blood glucose and eGFR by CKD-EPI eGFRCYSTATIN C over 6-years in the Coronary Artery Calcification in Type 1 diabetes (CACTI) study. All subjects with T1D and complete data including blood glucose and cystatin C for at least one of the three visits (n = 616, 554, and 521, respectively) were included in the longitudinal analyses.
In mixed-models adjusting for sex, HbA1c, ACEi/ARB, protein and sodium intake positive associations were observed between simultaneous blood glucose and eGFRCYSTATIN C (β±SE:0.14±0.04 per 10 mg/dL of blood glucose, p<0.0001), and hyperfiltration as a dichotomous outcome (OR: 1.04, 95% CI: 1.01–1.07 per 10 mg/dL of blood glucose, p = 0.02).
In our longitudinal data in subjects with T1D, simultaneous blood glucose has an independent positive effect on eGFRCYSTATIN C. The associations between blood glucose and eGFRCYSTATIN C may bias the accurate detection of early diabetic nephropathy, especially in people with longitudinal variability in blood glucose.
Citation: Bjornstad P, McQueen RB, Snell-Bergeon JK, Cherney D, Pyle L, Perkins B, et al. (2014) Fasting Blood Glucose-A Missing Variable for GFR-Estimation in Type 1 Diabetes? PLoS ONE9(4): e96264. https://doi.org/10.1371/journal.pone.0096264
Editor: Nick Ashton, The University of Manchester, United Kingdom
Received: November 26, 2013; Accepted: April 6, 2014; Published: April 29, 2014
Copyright: © 2014 Bjornstad et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Funding: Support for this study was provided by NHLBI grant R01 HL61753, HL79611, DERC Clinical Investigation Core P30 DK57516 ,and JDRF grant 17-2013-313. The study was performed at the Adult CTRC at UCD support by NIH-M01-RR00051, at the Barbara Davis Center for Childhood Diabetes and at Colorado Heart Imaging Center in Denver, CO. Dr. Snell-Bergeon was supported by an American Diabetes Association Junior Faculty Award (1-10-JF-50). Dr. McQueen was supported by a Post-Doctoral Grant from the PhRMA Foundation. D.Z.I.C. was supported by a Kidney Foundation of Canada Scholarship and a Canadian Diabetes Association-KRESCENT Program Joint New Investigator Award. Drs. Bjornstad, McQueen, Snell-Bergeon, Cherney, Perkins, Rewers and Maahs have no conflict of interest to disclose. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Competing interests: The authors have declared that no competing interests exist.
Diabetic nephropathy (DN) remains the single most important cause of renal failure in North America, and also one of the major causes of mortality in type 1 diabetes , . The American Diabetes Association, National Kidney Foundation and International Society of Nephrology recommend annual screening for albuminuria and also measurement of estimated glomerular filtration rate (eGFR) to identify and monitor DN –. Several factors influencing GFR are well documented and commonly accounted for when employing estimating equations to assess renal function including age, gender, and ethnicity . The effect of hyperglycemia on renal physiology and GFR measurement is well recognized in controlled studies , , but is not accounted for in GFR estimating equations, which could introduce bias in the 1.5 million patients in the US and 20 million worldwide with type 1 diabetes (T1D) –. Moreover, although eGFR equations have included subjects with T1D , only diabetes status was considered and not blood glucose concentrations as potential explanatory variables.
The influence of simultaneous blood glucose on eGFR (using serum creatinine) was suggested by the DCCT-EDIC study which showed a decrease of 4.2 mL/min/1.73 m2 in eGFR from baseline to year one among subjects randomized to intensive diabetes control and a sustained lower eGFR during the DCCT; similarly, eGFR decreased 4.8 mL/min/1.73 m2 from DCCT close-out to year 1 of EDIC in subjects in the conventional arm after transition to more intensive diabetes management . Furthermore, over 6 ½ years in the DCCT the overall difference in mean daytime blood glucose between intensively and conventionally treated subjects was 76 mg/dL and associated with a 1.7 mL/min/1.73 m2 overall increase in eGFR in the conventionally treated group , . One hypothesis to explain these findings is the effect of blood glucose control at the time of eGFR measurement.
Compared to creatinine-based equations, we have previously demonstrated that cystatin C has a superior ability to experimentally detect acute changes in GFR induced by hyperglycemia under carefully controlled physiological conditions . However, the role of simultaneously measured glucose on eGFR calculated by using cystatin C equations in people with T1D outside of a controlled laboratory setting is not known. It is important to determine the magnitude of the effect of changes in plasma glucose on eGFR to potentially provide more precise GFR estimates in people with T1D.
Accordingly, our aim was to determine the longitudinal association of simultaneous fasting glucose on eGFR using cystatin C-based equations in the Coronary Artery Calcification in T1D (CACTI) study, a large cohort of adults with T1D, over 6 years to better understand how simultaneously measured blood glucose affects eGFR over time. We hypothesized that eGFR would be positively associated with simultaneously measured blood glucose level. Second, and consistent with previous work we hypothesized that the association between glucose and eGFR would be weaker in patients taking RAAS inhibition (RAASi) than in those without RAASi, since RAASi attenuates the hemodynamic effects of hyperglycemia .
Materials and Methods
The study was approved by the Colorado Multiple Institutional Review Board and all participants provided written informed consent to participate in this study.
Cohort and Methods
The CACTI Study enrolled subjects 19–56 years old, with and without T1D, who were asymptomatic for cardiovascular disease (CVD) at the baseline visit in 2000–2002 and then were re-examined 3 and 6 years later, as previously described . Subjects with T1D had to have diabetes duration of at least 10 years at enrollment, with the exception of 13 subjects with a shorter duration at baseline who had taken part in a pilot study in 1997–1998 and were grandfathered in to the cohort. Subjects with serum creatinine >2 mg/dL were excluded at baseline, unless they were participants in the pilot study.
Study participants who completed the baseline screening visit were asked to fill out a validated self-administered food-frequency questionnaire, from which we obtained sodium and protein intake . We measured height and weight, and calculated body mass index (BMI) in kg/m2. Resting systolic (SBP) and fifth-phase diastolic blood pressure (DBP) were measured three times while the patient was seated, and the second and third measurements were averaged. Hypertension was defined as current anti-hypertensive therapy or untreated hypertension (BP ≥140/90 mmHg) at the time of the study visit. Anti-hypertension medication use was determined by a medication inventory as previously described  and use of an ACE inhibitor (ACEi) or an angiotensin receptor blocker (ARB) were combined for these analyses.
After an overnight fast, blood was collected, centrifuged, and separated. Blood glucose was measured using standard enzymatic methods in the laboratory and high performance liquid chromatography was used to measure HbA1c (HPLC, BioRad variant). Serum uric acid (SUA) was measured on stored baseline samples via the Clinical Analyzer utilizing a uricase-based commercial kit. These samples had been thawed twice in the past. The results were reported in milligrams per deciliter (mg/dL). LDL cholesterol (LDL-C) was calculated using the Friedewald formula. Timed overnight urine samples were collected and urine creatinine and albumin were measured (RIA, Diagnostic Products) at all three visits. Urinary albumin excretion rate (AER) was calculated and the results were reported in microgram per min (µg/min). Cystatin C was measured in the University of Colorado Hospital clinical lab using the commercially available Dade-Behring assay following package insert instructions on a BNII or Prospec instrument, as previously described in detail . Due to a systematic shift in the Dade-Behring cystatin C assay over the time period of our study, cystatin C levels were standardized to Visit 3 levels using Deming regression equations as previously described . GFR (ml/min/1.73 m2) was determined using the CKD-EPI eGFRCYSTATIN C, equation recently published by the CKD-EPI Investigators Group .
Categorical analyses were also performed based on eGFR and blood glucose to investigate whether thresholds of hyperfiltration affected these associations. There is no generally accepted definition for hyperfiltration . In our study, we determined sex-specific reference values for hyperfiltration using the upper 90th percentile of eGFR from normal controls of CACTI to define hyperfiltration (eGFR >120 mL/min/1.73 m2 reflected approximately the 90th percentile for both male and female controls by CKD-EPI eGFRCYSTATIN C).
Differences between variables were assessed using paired t-tests. AER had a non-parametric distribution and were presented as geometric means and natural log-transformed in the multivariable mixed models. To explore the relationships between blood glucose and eGFR as continuous outcome and categorical outcome (hyperfiltration, eGFR > 120 mL/min/1.73 m2) at all three visits and how these relationships differ over time, we employed longitudinal linear and logistic mixed-models , . These models strengthen the cross-sectional evidence by observing the same associations over repeated time points. To investigate whether the associations between glucose and eGFR and hyperfiltration were independent of confounding variables, we adjusted for sex, sodium intake, protein intake, HbA1c, ACEi/ARB use. We adjusted for HbA1c, as we sought to explore the effects of acute glucose on eGFR rather than long term glycemic control which could confound our results.
The mixed model written in observation-specific form is shown below:
i = 1, 2, 3, …, n subjects
j = 1, 2, 3 measurements at each time period or visit
Spot glucose (mg/dL), HbA1c (%), sodium intake, protein intake, and time were treated as continuous variables. Female was coded as dichotomous with male as the reference group. ACEi/ARB was coded as dichotomous with no ACEi/ARB use as the reference group. is a random intercept representing the individual intercept deviation. is a random slope on time representing the individual slope deviation for time. The random effects have mean 0 and we assumed an unstructured covariance matrix to allow for correlation between the random effects. The interaction term between glucose and time represents the per time period change in spot glucose on eGFR levels (i.e., how does the slope vary across time). The reported association between glucose and eGFR is a marginal effect calculated as , where is the mean observed time for participants across all three visits; similar to a pooled estimate across all visits. The marginal effect represents the average difference in eGFR for every 10-unit difference in glucose (e.g., 10 mg/dL for blood glucose), after adjusting for confounding variables. A similar approach was taken for the logistic mixed models using hyperfiltration as a dichotomous outcome, with the following abbreviated specification:
Note that is the same vector of covariates included in the linear mixed model. The model with the best fit statistics included only a random intercept term, , assumed to follow a normal distribution with mean 0 and variance . The marginal effect reported is also a function of glucose, and the interacted term between glucose and time. However, the interpretation using mixed effects logistic regression is subject-specific (i.e., conditional on the random effect). That is, for a given patient, there was an x% change in the odds of hyperfiltration for every 10-unit difference in glucose (e.g., 10 mg/dL for blood glucose), after adjusting for confounding variables. Gender and RAASi have both been suggested to influence blood glucose's effect on eGFR, and therefore we present non-stratified models in addition to models stratified by gender and ACEi/ARB therapy. We attempted multiple sensitivity analyses including different model and random effects specifications. We also performed post-hoc sensitivity analyses adjusting for BMI, LDL-C, SUA at baseline, AER and SBP along with the existing variables (sex, HbA1c, ACEi/ARB usage, protein intake and sodium intake) to further assess the relationship between blood glucose and eGFRCYSTATIN C and hyperfiltration. The Akaike Information Criteria (AIC), log-likelihood, and residual plots were used to analyze goodness of fit. All analyses were performed in Stata (Stata Statistical Software: Release 12. College Station, TX: StataCorp LP) using the xtmixed and xtlogit commands. A value of P < 0.05 was considered statistically significant.
The characteristics of subjects with T1D at visit 1, 2 and 3 are summarized in Table 1. In non-stratified mixed-models adjusting for sex, HbA1c, ACEi/ARB usage, protein intake and sodium intake a positive association was observed between simultaneously measured blood glucose and eGFRCYSTATIN C (β±SE: 0.14±0.04, p<0.0001, per 10 mg/dl of blood glucose, Figure 1). Stratified by gender the associations between blood glucose and eGFRCYSTATIN C remained significant in both females (β±SE: 0.15±0.05, p = 0.003) and males (β±SE: 0.14±0.06, p = 0.03, Table 2). Moreover, stratified by ACEi/ARB use, subjects without RAASi demonstrated a significant positive association between blood glucose and eGFRCYSTATIN C (β±SE: 0.14±0.05, p = 0.005), in contrast to a non-significant association in those with RAASi (β±SE:0.13±0.07, p = 0.07, Table 2).
In non-stratified mixed-models with hyperfiltration as a dichotomous outcome, glucose was associated with hyperfiltration (OR = 1.04 [95% CI: 1.01–1.07], p = 0.02) in a fully-adjusted model. This translates to a 4% increase in odds of having hyperfiltration after multivariable-adjustments for every 10 mg/mL higher blood glucose. Stratified by ACEi/ARB therapy, the association between glucose and hyperfiltration was only significant for subjects without RAASi (ACEi/ARB use: OR = 1.02 [95% CI: 0.96–1.09], p = 0.43, no ARBi/ARB use: OR = 1.04 [95% CI: 1.03–1.09], p = 0.03) (Figure 2).
To further assess the relationship between blood glucose and eGFRCYSTATIN C and hyperfiltration we performed post-hoc sensitivity analyses adjusting for BMI, LDL-C, SUA at baseline, AER and SBP along with the existing variables (sex, HbA1c, ACEi/ARB usage, protein intake and sodium intake). In non-stratified models glucose remains significantly associated with eGFR CYSTATIN C and hyperfiltration (Table 3). Stratified by ACEi/ARB use, subjects without RAASi demonstrated a significant positive association between blood glucose and eGFRCYSTATIN C and hyperfiltration, in contrast to non-significant associations with eGFRCYSTATIN C and hyperfiltration in those with RAASi (Table 3).
Our major findings in this descriptive study were that simultaneously measured blood glucose was positively associated with eGFR and with hyperfiltration calculated by eGFRCYSTATIN C in subjects with T1D after multivariable adjustments, which is consistent with experimental data  and with our hypothesis. The rationale for our study was twofold: 1) simultaneous hyperglycemia is known to affect GFR under experimental conditions , –; and 2) while simultaneous hyperglycemia may increase GFR under controlled settings, the association of blood glucose variability with cystatin C based eGFR measurements is uncertain in patients with T1D in clinical studies or clinical care settings.
This question is clinically important as eGFR and microalbuminuria are the only two methods recommended by American Diabetes Association, National Kidney Foundation and International Society of Nephrology to screen for and monitor progression of DN in the clinical setting –. As microalbuminuria does not necessarily lead to macroalbuminuria, and in fact may regress spontaneously , , estimation of GFR has an important role in risk stratification. Accordingly, clinicians rely upon calculations of eGFR to identify early DN (e.g. rapid GFR decline [annual loss >3mL/min/1.73 m2]) and risk-stratify subjects with T1D –. Furthermore, a recent report by Krolewski et al. shows that rapid GFR decline occurs prior to the onset of microalbuminuria  supporting its role in early risk stratification. Recent data from the DCCT-EDIC study suggested that blood glucose at the time of eGFR measurement may bias results, as subjects randomized to intensive diabetes control displayed a lower eGFR at one year in the DCCT and also in EDIC among those in the DCCT conventional arm who then intensified their blood glucose control .
The etiology of increased GFR in the setting of elevated blood glucose in subjects with T1D is incompletely understood, but has been attributed to the effect of hyperglycemia on RAAS. Miller et al. demonstrated that hyperfiltration responses to clamped hyperglycemia are related to intrarenal RAAS activation , . Moreover, Cherney et al. recently showed that RAAS blockade by aliskiren (a direct renin inhibitor) blunts the increased GFR as measured by inulin clearance provoked by hyperglycemia . These findings are also consistent with our data, where we did not observe a significant association between glucose and eGFRCYSTATIN C in subjects with RAAS inhibition. Furthermore, hyperglycemia has been proposed to increase proximal tubular glucose delivery causing a maladaptive increase in glucose reabsorption along with sodium via sodium-glucose cotransporter 2 (SGLT-2) in the proximal tube. Consequently – distal sodium chloride delivery to the macula densa is decreased. This decrease is perceived as low effective circulating volume by the juxtaglomerular apparatus, which causes vasodilation of the afferent renal arteriole and an increased GFR . There is to our knowledge no current evidence of a direct effect of glucose on cystatin C, suggesting the positive relationship between blood glucose and eGFRCYSTATIN C is likely reflective of changes in GFR.
Previous experimental studies to date suggest a positive relationship between fasting blood glucose and measured GFR, an association which is consistent with the relationship between simultaneously measured blood glucose and eGFRCYSTATIN C in our study. The CKD-EPI equations estimate GFR, but have increased bias and less precision when GFR is >60 mL/min/1.73 m2 . Previous experimental work has consistently demonstrated that serum creatinine-based methods cannot detect acute changes in GFR , . The literature is less consistent with regards to cystatin C . Cherney et al reported that GFR measured by inulin clearance results in a GFR difference of 15–18 mL/min/1.73 m2 when blood glucose is clamped at 4–6 mmol/L vs. 9–11 mmol/L in adult patients with T1D . Moreover, they demonstrated a strong agreement between eGFRCYSTATIN C and GFRINULIN in detecting increases in GFR provoked by clamped hyperglycemia  in subjects with T1D, but this relationship was not evident using eGFRCREATININE. In contrast, Melsom et al did not demonstrate statistical agreement between eGFRCYSTATIN C and GFRIOHEXOL in detecting GFR changes, however this was in non-diabetic individuals with impaired fasting glucose in whom there is less blood glucose variability than in people with T1D .
The utility of cystatin C as a marker of renal function versus creatinine based estimates is controversial, as reviewed elsewhere . Cystatin C was used in this study instead of creatinine due to its superior operating characteristics as a measure of renal function changes in response to ambient glycemia. eGFRCYSTATIN C is considered to be less biased by age and weight compared to creatinine-based measurements . It is also recognized that blood glucose concentrations above 300 mg/dL may cause an overestimation of serum creatinine levels (and consequently an underestimation of eGFRCREATININE) , . Moreover, eGFRCYSTATIN C appears to predict micro- and macrovascular complications in subjects with T1D better than eGFRCREATININE , , . Skupien et al. also recently demonstrated that GFR staging with eGFRCYSTATIN C is superior for predicting ESRD and mortality than staging by the combined eGFR creatinine and cystatin C equation, which they hypothesized might suggest that some determinant of serum creatinine counters the predictive effect of serum cystatin C . Finally, Shlipak et al. recently demonstrated that the use of eGFRCYSTATIN C compared to eGFRCREATININE strengthens the association between eGFR and risk of death and end-stage renal disease in 11 diverse general-population studies .
An important question derived from our data is whether the statistically significant associations observed between simultaneously measured blood glucose and eGFRCYSTATIN C is clinically significant and warrants systematic adjustment for blood glucose when employing these equations to estimate GFR in subjects with T1D. In fully adjusted models, an increase in simultaneously measured blood glucose of 10 mg/dL would be associated with an eGFRCYSTATIN C increase of 0.14 mL/min/1.73 m2, or 1.4 mL/min/1.73 m2 for an increase of 100 mg/dL. Glycemic excursions greater than 100 mg/dL are not uncommon in people with type 1 diabetes. With rapid GFR decline defined as an annual loss >3 mL/min/1.73 m2 or >3.3% , , such excursions could significantly bias the detection of early DN. For that reason, adding blood glucose to eGFR estimation in people with T1D might improve intra-individual precision over time, which could ultimately improve screening, detection, and prevention of early changes in the GFR . Early identification of subjects at risk of DN is imperative to prevent early morbidity and mortality .
Experimentally, the association between blood glucose and measured GFR has been well described , , , but this is to our knowledge the first study exploring the longitudinal relationships between simultaneously measured fasting blood glucose and eGFR, measured by CKD-EPI cystatin C, in a large clinical cohort of subjects with T1D. Nevertheless, there are important limitations to the present study worth mentioning, including the observational design. Moreover, experimental studies can control blood glucose prior to measuring GFR, whereas we had to rely on simultaneously measured fasting blood glucose to represent blood glucose control in the period prior to eGFR measurement, which is known to affect renal hemodynamics. The blood glucose samples in our study were drawn from fasting subjects, and therefore the variability in blood glucose would likely be greater in non-fasting samples. Another limitation was the use of cystatin C based equation rather than direct measurements of GFR, which would have been too cumbersome for use in a large-scale clinical study like CACTI. Without the comparison of a direct measure of GFR we were unable to create a cystatin C based eGFR algorithms adjusting for glucose. More importantly our main goal was to elucidate the effects of simultaneously measured blood glucose on the cystatin C-based estimating equation that are commonly used in research in people with T1D. Although we adjusted for a variety of important confounding variables, we cannot rule out the presence of other risk factors that may have biased the present analyses. However, the purpose of our analyses was not to test the independence relationship between blood glucose and GFR as this is well recognized, but rather elucidate the magnitude of the relationship between blood glucose and eGFRCYSTATIN C outside of the laboratory setting.
In summary, we report a positive independent effect of simultaneously measured blood glucose on eGFR and hyperfiltration by eGFRCYSTATIN C in subjects with T1D over time. The magnitude of the β-coefficient between glucose and eGFRCYSTATIN C may be considered of modest clinical significance. In contrast, in subjects with T1D who experience significant glycemic excursions, the magnitude of the associations may bias the detection of early DN. Further study with measured GFR is required to determine whether systematic adjustments for simultaneously measured glucose in GFR equations for adults with T1D are warranted.
Conceived and designed the experiments: PB RBM JKS DC LP BP MR DMM. Performed the experiments: PB RBM JKS DC LP BP MR DMM. Analyzed the data: PB RBM JKS DC LP BP MR DMM. Wrote the paper: PB RBM JKS DC LP BP MR DMM.
- 1. Maahs DM, Rewers M (2006) Editorial: Mortality and renal disease in type 1 diabetes mellitus—progress made, more to be done. J Clin Endocrinol Metab 91: 3757–3759.
- 2. Orchard TJ, Secrest AM, Miller RG, Costacou T (2010) In the absence of renal disease, 20 year mortality risk in type 1 diabetes is comparable to that of the general population: a report from the Pittsburgh Epidemiology of Diabetes Complications Study. Diabetologia 53: 2312–2319.
- 3. American Diabetes Association (2013) Standards of medical care in diabetes—2013. Diabetes Care 36 Suppl 1S11–66.
- 4. National Kidney Foundation (2007) KDOQI Clinical Practice Guidelines and Clinical Practice Recommendations for Diabetes and Chronic Kidney Disease. Am J Kidney Dis 49: S12–154.
- 5. Stevens PE, Levin A (2013) Evaluation and management of chronic kidney disease: synopsis of the kidney disease: improving global outcomes 2012 clinical practice guideline. Ann Intern Med 158: 825–830.
- 6. Inker LA, Schmid CH, Tighiouart H, Eckfeldt JH, Feldman HI, et al. (2012) Estimating glomerular filtration rate from serum creatinine and cystatin C. N Engl J Med 367: 20–29.
- 7. Wiseman MJ, Saunders AJ, Keen H, Viberti G (1985) Effect of blood glucose control on increased glomerular filtration rate and kidney size in insulin-dependent diabetes. N Engl J Med 312: 617–621.
- 8. Perkins BA, Sochett EB, Cherney DZ (2012) Ability of Cystatin C to detect changes in glomerular filtration rate after ACE inhibition in patients with uncomplicated type 1 diabetes. Clin Exp Hypertens 34: 606–611.
- 9. Danaei G, Finucane MM, Lu Y, Singh GM, Cowan MJ, et al. (2011) National, regional, and global trends in fasting plasma glucose and diabetes prevalence since 1980: systematic analysis of health examination surveys and epidemiological studies with 370 country-years and 2.7 million participants. Lancet 378: 31–40.
- 10. Lervang HH, Jensen S, Brochner-Mortensen J, Ditzel J (1988) Early glomerular hyperfiltration and the development of late nephropathy in type 1 (insulin-dependent) diabetes mellitus. Diabetologia 31: 723–729.
- 11. Yip JW, Jones SL, Wiseman MJ, Hill C, Viberti G (1996) Glomerular hyperfiltration in the prediction of nephropathy in IDDM: a 10-year follow-up study. Diabetes 45: 1729–1733.
- 12. de Boer IH, Sun W, Cleary PA, Lachin JM, Molitch ME, et al. (2011) Intensive diabetes therapy and glomerular filtration rate in type 1 diabetes. N Engl J Med 365: 2366–2376.
- 13. The Diabetes Control and Complications Trial Research Group (1993) The effect of intensive treatment of diabetes on the development and progression of long-term complications in insulin-dependent diabetes mellitus. The Diabetes Control and Complications Trial Research Group. N Engl J Med 329: 977–986.
- 14. Cherney DZ, Sochett EB, Dekker MG, Perkins BA (2010) Ability of cystatin C to detect acute changes in glomerular filtration rate provoked by hyperglycaemia in uncomplicated Type 1 diabetes. Diabet Med 27: 1358–1365.
- 15. Miller JA (1999) Impact of hyperglycemia on the renin angiotensin system in early human type 1 diabetes mellitus. J Am Soc Nephrol 10: 1778–1785.
- 16. Maahs DM, Kinney GL, Wadwa P, Snell-Bergeon JK, Dabelea D, et al. (2005) Hypertension prevalence, awareness, treatment, and control in an adult type 1 diabetes population and a comparable general population. Diabetes Care 28: 301–306.
- 17. Snell-Bergeon JK, Chartier-Logan C, Maahs DM, Ogden LG, Hokanson JE, et al. (2009) Adults with type 1 diabetes eat a high-fat atherogenic diet that is associated with coronary artery calcium. Diabetologia 52: 801–809.
- 18. Maahs DM, Jalal D, McFann K, Rewers M, Snell-Bergeon JK (2011) Systematic shifts in cystatin C between 2006 and 2010. Clin J Am Soc Nephrol 6: 1952–1955.
- 19. Sunder-Plassmann G, Horl WH (2004) A critical appraisal for definition of hyperfiltration. Am J Kidney Dis 43: 396; author reply 396–397.
- 20. Hedeker D, Gibbons RD (2006) Longitudinal Data Analysis: John Wiley & Sons.
- 21. Cameron CA, Triverdi PK (2010) Microeconometrics Using Stata: STATA Press.
- 22. Yang GK, Maahs D, Perkins BA, Cherney DZ (2013) Renal hyperfiltration exaggerates the renin angiotensin aldosterone system paradox in patients with type 1 diabetes. In Review.
- 23. Cherney DZ, Miller JA, Scholey JW, Nasrallah R, Hebert RL, et al. (2010) Renal hyperfiltration is a determinant of endothelial function responses to cyclooxygenase 2 inhibition in type 1 diabetes. Diabetes Care 33: 1344–1346.
- 24. Cherney DZ, Scholey JW, Sochett EB (2013) Sex differences in renal responses to hyperglycemia, L-arginine, and L-NMMA in humans with uncomplicated type 1 diabetes. Diabetes Care 36: 1290–1296.
- 25. Perkins BA, Krolewski AS (2009) Early nephropathy in type 1 diabetes: the importance of early renal function decline. Curr Opin Nephrol Hypertens 18: 233–240.
- 26. Perkins BA, Ficociello LH, Ostrander BE, Silva KH, Weinberg J, et al. (2007) Microalbuminuria and the risk for early progressive renal function decline in type 1 diabetes. J Am Soc Nephrol 18: 1353–1361.
- 27. Shlipak MG, Katz R, Kestenbaum B, Fried LF, Newman AB, et al. (2009) Rate of kidney function decline in older adults: a comparison using creatinine and cystatin C. Am J Nephrol 30: 171–178.
- 28. Shlipak MG, Katz R, Kestenbaum B, Siscovick D, Fried L, et al. (2009) Rapid decline of kidney function increases cardiovascular risk in the elderly. J Am Soc Nephrol 20: 2625–2630.
- 29. Bjornstad P, Snell-Bergeon JK, Rewers M, Jalal D, Chonchol MB, et al.. (2013) Early Diabetic Nephropathy: A complication of reduced insulin sensitivity in type 1 diabetes. Diabetes Care.
- 30. Krolewski AS, Niewczas MA, Skupien J, Gohda T, Smiles A, et al. (2014) Early progressive renal decline precedes the onset of microalbuminuria and its progression to macroalbuminuria. Diabetes Care 37: 226–234.
- 31. Miller JA, Floras JS, Zinman B, Skorecki KL, Logan AG (1996) Effect of hyperglycaemia on arterial pressure, plasma renin activity and renal function in early diabetes. Clin Sci (Lond) 90: 189–195.
- 32. Cherney DZ, Reich HN, Scholey JW, Daneman D, Mahmud FH, et al. (2013) The effect of aliskiren on urinary cytokine/chemokine responses to clamped hyperglycaemia in type 1 diabetes. Diabetologia 56: 2308–2317.
- 33. Cherney DZ, Perkins BA, Soleymanlou N, Maione M, Lai V, et al.. (2013) The Renal Hemodynamic Effect of SGLT2 Inhibition in Patients with Type 1 Diabetes. Circulation.
- 34. Melsom T, Mathisen UD, Ingebretsen OC, Jenssen TG, Njolstad I, et al. (2011) Impaired fasting glucose is associated with renal hyperfiltration in the general population. Diabetes Care 34: 1546–1551.
- 35. Odutayo A, Cherney D (2012) Cystatin C and acute changes in glomerular filtration rate. Clin Nephrol 78: 64–75.
- 36. Husdan H, Rapoport A (1968) Estimation of creatinine by the Jaffe reaction. A comparison of three methods. Clin Chem 14: 222–238.
- 37. Silveiro SP, Araujo GN, Ferreira MN, Souza FD, Yamaguchi HM, et al. (2011) Chronic Kidney Disease Epidemiology Collaboration (CKD-EPI) equation pronouncedly underestimates glomerular filtration rate in type 2 diabetes. Diabetes Care 34: 2353–2355.
- 38. Maahs DM, Ogden LG, Kretowski A, Snell-Bergeon JK, Kinney GL, et al. (2007) Serum cystatin C predicts progression of subclinical coronary atherosclerosis in individuals with type 1 diabetes. Diabetes 56: 2774–2779.
- 39. Premaratne E, MacIsaac RJ, Finch S, Panagiotopoulos S, Ekinci E, et al. (2008) Serial measurements of cystatin C are more accurate than creatinine-based methods in detecting declining renal function in type 1 diabetes. Diabetes Care 31: 971–973.
- 40. Skupien J, Warram JH, Groop PH, Krolewski AS (2013) Cystatin-based estimated GFR versus creatinine-based and creatinine- and cystatin-based estimated GFR for ESRD and mortality risk in diabetes. Am J Kidney Dis 62: 184–186.
- 41. Shlipak MG, Matsushita K, Arnlov J, Inker LA, Katz R, et al. (2013) Cystatin C versus creatinine in determining risk based on kidney function. N Engl J Med 369: 932–943.
- 42. Cherney D, Maahs DM (2013) Cystatin C versus creatinine for kidney function-based risk. N Engl J Med 369: 2458.
- 43. Greene SA, Dalton RN, Turner C, Haycock GB, Chantler C (1987) Hyperglycemia with and without glycosuria: effect on inulin and para-amino hippurate clearance. Kidney Int 32: 896–899.